This project is a demo of an LLM-based interactive technical/functional specification writing tool. It is a web application that helps developers, product managers, and planners create MVP technical specifications more efficiently.
- Support non-technical planners and business stakeholders to easily organize technical and business requirements when writing service/product planning documents, as if working with a developer for technical validation.
- Present an optimal document format that can reflect technical requirements suitable for agent-driven development methodologies such as Cursor and Windsurf.
- AI-powered conversational interface (Amazon Bedrock 3.7 Sonnet)
- User authentication via AWS Cognito
- Web search functionality (Tavily API integration)
- Asynchronous collaboration through chat session sharing
- Python 3.13 or higher
- Node.js 20 or higher
- Docker
- AWS CLI configuration or Claude API Key
alps-writer/
├── packages/
│ ├── app/ # Main application code
│ └── infra/ # AWS CDK infrastructure code
└── docs/ # Project documentation
For detailed installation and setup instructions for each package, refer to the README.md in the respective directory:
You can run this project either locally on your machine:
Or deploy it to your personal AWS account:
The following features are planned for production-ready release:
- Apply RAG and chat-history-memory features to reduce token usage
- Enable Prompt Caching for reducing token usage
- TaskManager feature that reduce blast radius during develop using Agentic IDEs
- Add Save command for upload completed document to S3 with one click
- Migrate from Chainlit to custom solution that supports concurrent user sessions per each conversation
- Build document management system for specification generation and versioning
Apache License 2.0